This chapter will briefly review the research questions addressed in the study and the accompanying results. The limitations in the study are presented so the results are placed in proper context. Implications for research and practice will be reviewed before closing remarks.
Overview of this Research
The primary purpose of this research was the clarification of qualities of the psychological construct of value. Three research questions were asked related to within subject stability of value across contexts, the stability of value ratings across subjects in different domains of interest, and the relationship between perceived value and perceived regret.
Within subject stability for hypothesis one was measured through the rating of the same stimulus items by the participant at two different time points and under two
different conditions. A paired samples t-test was the primary assessment measure used and indicated that any difference in ratings from Time One to Time Two could not be statistically attributed to chance.
To determine if there was a difference in stability ratings between the different categories of stimulus items presented, absolute value difference scores were created for each participant. The difference scores for all participants were averaged for the
60
monetary stimulus item category, the tangible stimulus item category, and the intangible stimulus item category. Paired sample t-tests were run for each pairing of stimulus item category: monetary and tangible, monetary and intangible, and tangible and intangible. Statistically significant differences were discovered in all three pairings of stimulus item categories.
The relationship between perceived regret and perceived value in a choice
condition was hypothesized to follow a specified pattern. Participants made value ratings during a choice condition that requested an additional regret rating. Based on the choices and ratings made by the participants, each instance of a choice following the expected pattern was considered true and each instance that did not follow the pattern was
considered false. A binomial distribution comparison could then be made to consider if the number of responses found true out of all choice conditions was statistically
significant for that participant. After 44 binomial distribution comparisons, the participants could be compared to a binomial distribution for number of participants found to be statistically significant out of the total number of participants. A statistically significant number of participants were found to follow the hypothesized pattern of value and regret ratings.
Discussion of Results
The results of these three hypotheses suggest that there may be strong, stable, and predictable qualities related to the construct of value. Hypothesis one involved the question of stability. Though the literature base refers obliquely to the quality of stability in philosophy and culture (Schwartz & Bardi, 2001), motivation (Atkinson, 1957,
61
Gettis, & Ogden, 1999; Litt, Eliasmith, & Thagard, 2008; Sanfey, Loewenstein, McClure, & Cohen, 2006) it has not been the subject of direct empirical study. These results
indicate that 63.6% of participants (28 out of 44) showed stable value ratings on stimulus items from Time One to Time Two. This could indicate that the affective properties of stimulus items help to create valence and arousal memory traces that are stored for use in future decision making.
It is important to note that no comparison was made on an individual level for how someone rated an item compared to someone else. These ratings, and the stability results, are only personal and individual ratings and scores.
Hypothesis two raised the question of stability of value ratings between different domains. The results of significant differences between the absolute value difference score ratings of the three domains of monetary (M = .62), intangible (M = 1.64), and tangible (M = 1.93) stimuli categories is not surprising and mirrors much of the work presented on preferences, value, and reward between money (Knutson, Adams, Fong, & Hommer, 2001), food items (McClure, Li, Tomlin, Cypert, & Montegue, 2004), and social situations (Rilling, Sanfey, Aronson, Nystom, & Cohen, 2004). Preferences are believed to be affective and cognitive processes that fix specific valence and arousal patterns to specific items (Litt, Eliasmith, & Thagard, 2008). Variances should be expected between stimuli. The fact that the categories themselves showed such differences, and not just individual items, is not a finding in the existing research and opens another avenue for additional research in value domains.
This also serves as a cautionary note to future researchers. There is no reason to believe that an individual who has somewhat similar ratings in one field or domain will
62
have similar ratings in another domain. Domain specific research must be careful of genereralizations and multi-domain research in value must be aware of the potential differences across domains.
Hypotheis three investigated regret and value. Regret and value, though both integral parts of the decision process, have not been studied together in any explicit manner previously. The relationship between the two constructs is strong and predictable in this research with 88.6% of participants (39 out of 44) following the value and regret hypothesis. This suggests the possibility of using these two constructs together in future research to improve understanding of the decision process.
From a validity viewpoint, the greatest result presented is from hypothesis three. Regret is an incredibly stable construct that has been empirically studied in multiple domains using the same definition. The hypothesized relationship between value and regret was strong.
This process of measurement, using two constructs that are hypothetically linked in some way and determining if they are, indeed, related as hypothesized, is referred to as convergent validity (Trochim, 2006). This is but one part of the construct validity
process, but an important part from a research perspective. Replication of these results will provide researchers with strong evidence of a solid relationship between the two constructs that can then be used in discriminant and content validity efforts (Slavin 2007, Trochim, 2006).
Limitations
A number of limitations may impact the interpretation of these results. Value, as a construct, is considered to be universal in that all nations and peoples on the planet have
63
the capacity for value generation for use in individual or cultural judgment situations (Phinney, Ong, & Madden, 2000; Schwartz & Bardi, 2001). Considering the population possibilities, the sample of 44 undergraduate participants from a particular university in the southeastern United States is not sufficient to make any generalizations.
Additionally, the sample was very homogeneous in respect to cultural diversity. In addition, the students who did not complete the procedures and were not included in the study may have impacted the results. Listwise deletion, as a process, may potentially have an impact on statistical power. There is also a lack of data on the
students who did not complete the procedures leaving a question as to biases that may be inherent in the project that have gone unnoticed (Myers, 2011).
The within subject question of stability suffers from a number of methodological issues. Considering the need to address how an individual would rate the importance of an item multiple times, the balance between multiple ratings and the concern of a participant remembering a previous rating of a particular stimulus was delicate. A third rating of the same stimulus items may well have better served the question of stability, however, what time delay would be appropriate to ensure no previous memory of the ratings applied to the third round of stimulus ratings? One solution would be to maintain the 30 day delay in hopes that no memory consolidation of the previous experience was possible. It is also possible that a priming test of items similar to, but not exact repeats of, the target stimulus items may have interfered with memory consolidation. Another option would be to wait a period of time designated by existing research to be sufficient in avoiding memory consolidation of the target stimulus items.
64
Another empirical issue relates to the stimulus items used in all three research questions. The stimulus items were specifically selected to be representative of objects, situations, and scenarios that would prove common, or at least familiar, to the
participants. This may well have produced a cultural bias in the representative stimulus items. One method to validate or refute this concern would be to replicate the study with participants from multiple cultural backgrounds and compare mean and variance scores of the ratings. This may provide insight into what cultural biases may exist for each monetary, tangible, and intangible item.
Finally, another issue regarding the stimulus items used involves the method of presentation. As noted previously, images were either found or created for each stimulus item in the study and included on the cards along with explanatory text. As perception may be affected by affective response, so may interpretation of visual stimuli. As such, it is quite possible that the images used, particularly for the stimuli listed as intangible, made the item perceptually more tangible. This, in turn, could affect the rating on the stimulus item. With a sufficient number of subjects, a counter to this issue would be the division of the total sample into two equal groups both numerically and demographically. One group would go through all three procedures using stimulus cards as presented in this study. The second group would use stimulus cards that had no images, only text descriptions. Comparative analysis at the end of the study might then shed light on the potential impact the images may have provided.
One overarching and all-encompassing issue regards the very nature and definition of value. As previously stated, there are multiple ways in which value has been studied in the social sciences. Unfortunately, no common definition exists to guide
65
the individual and domain specific efforts toward a common goal of understanding. This study suffers from the same symptoms as the only truly definitional aspect of value included was the idea of importance and stability. The importance component was derived as synonymous for value use in previous studies. The stability component was implied in previous research and empirically tested here.
Implications for Future Research
Despite the positive experimental results and extensive literature, only the
simplest of approximations for a definition of value was used in the research. The reason is that there does not seem to be a consistent definition of value to guide further research into the construct. If one were to study value only in the field of motivation, the
definition most commonly used in that field could be used, but does that definition actually assist the construction of knowledge about the construct if another definition is operationalized in another field? A common definition of value, applicable to and studied within, multiple fields and domains of study is needed.
Without the common definition, and without established reference points from which to work, the majority of research in the realm of value tends to be from an
empirical perspective that is severely limited in its generalizability to other uses of value as a construct. As value has the potential to be vital to the understanding of human behavior, this issue needs to be addressed. As Pham, et. al. (2001) state, “It is not a coincidence that the valuation of objects – whether products, issues, or people – has always been a major subject of inquiry in all social sciences.” (p. 167)
This raises serious construct validity issues that would serve as an excellent starting point for further research on value. This study is the first step towards a common
66
definition, however additional research in this area is necessary and must address the definition and origin of value as a construct.
The beginning of the search for construct validity must begin with a definition. The importance of a strong, theoretically based, definition cannot be understated for ultimately, the measure created to gauge the construct will be based upon what the construct is believed to represent (Slavin, 2007; Smith, 2005; Strauss & Smith, 2009). The definition of the construct, therefore, informs how the construct interacts with other constructs or processes and helps to define the measures that will be used to help
determine its existence.
Trochim (2006) uses the phrase, “Inadequate Preoperational Explication of Constructs,” to help elucidate definitional issues. In this, he states that, even before the construct has been placed into an operational conception, that is, before the researcher attempts to determine how the construct will actually interact with other constructs, processes, and individuals, one must be especially careful to fully conceptually define the construct. Failure to do this is the first threat to construct validity. Whether the
definitional process involves concept mapping, extensive research, or the assistance of other researchers in the field, the first step of construct validity must be a thorough investigation of the proposed construct (Trochim, 2006).
It should be noted that the idea of construct validity subsumes the ideas of content validity, criterion-related validity, and other validity types related specifically to
psychological constructs, is well established (Slavin, 2007; Smith, 2005; Strauss & Smith, 2009; Trochim, 2006; Westen & Rosenthal, 2003).
67
The issues with the construct of value, as illustrated by the research in different fields of study, point to one central idea: the definition of a construct must be well formed if it is to meet the requirements of the researchers who use said construct. The current evidence suggests that the construct of value has very little common ground for assessment and has no grounds for generalizability: hence, no construct validity. Therefore, the creation of a definition must be undertaken with exceptional care and diligence.
Based on the existing literature from across multiple disciplines that have used the construct of value as a variable in empirical studies, a definition for value that may serve as an opening effort into the true investigation of the construct follows: value is an evolutionarily conserved biological process, affective in nature, that exhibits the properties of power, stability, and cognitive interaction.
Justification for Future Research
To justify future research into the definition of value, the components of the definition, how they tie into the existing literature, and how the current studies may contribute to their inclusion are discussed, beginning with a discussion of affect and suggesting that affect is biological. Future research possibilities for an evolutionarily based origin of affect will link to existing literature and research studies.
One definition of affect, provided by Frijda and Scherer (2009) in The Oxford Companion to Emotion and the Affective Sciences, states, “The terms affect and the corresponding adjective affective are generally used in an overarching generic sense for a mental state that is characterized by emotion feeling as compared with rational thinking” (p. 10). The definition continues to offer more
68
detail by elaborating with, “the term is often used in a general sense to refer to a class or category of mental states that includes emotions, moods, attitudes, interpersonal stances, and affect dispositions” (p. 10). Another definition for affect provided by Zeelenberg, et. al. (2008) posits, “Affect is a generic term that refers to many experiential concepts including moods, emotions, attitudes, evaluations and preferences” and continues with, “Thus, any experiential concept that is positive or negative can be considered affective” (p. 19).
Connected to affect, then, are personal representations of what is good or bad. Cunningham and Zelazo (2007) indicate that the term valence represents whether a stimulus is perceived as good or bad. Pham, Cohen Pracejus, and Hughes (2001) state that arousal is directly related to the intensity of the valence a particular stimulus represents. Together, valence and arousal create an appraisal process that determines whether an individual views a particular stimulus positively or negatively, and how positively or how negatively. This conception is repeatedly supported throughout the affect literature. In the affect-as-information literature, Storbeck and Clore (2008) are as clear as possible by stating that, “Affective experience is often treated as having two dimensions – valence and arousal” (p. 1862). The terms arise again with Panksepp’s (2005) discussion of affective consciousness as the traditional views of affective mental representation. Davidson (2003); Deak (2011); Desmeules, Bechara, and Dube (2007); Izard (2010); Litt, Eliasmith, and Thagard (2008); Murphy, Nimmo-Smith, and Lawrence (2003); Posner, et. al. (2009); and Zeelenberg, Nelissen, Breugelmans, and Pieters (2008) are all published researchers in the field of affect, or affective related issues, who use valence and arousal as core assumptions in their understanding.
69
Why should affect, or emotion, be included in the definition for value? The central feature of the motivational and decision making research is that individuals will act in such a way as to maximize a valued goal or end state (Nagengast, Marsh, & Scalas, 2011; Shah & Higgins, 1997). The focus of achievement motivation research to
“…explain people’s choice of achievement tasks, persistence on those tasks, vigor in carrying them out, and performance on them” (Wigfield & Eccles, 2000) is directly tied to how desirable the end state of those tasks is viewed. The level of desirability is the valence and arousal appraisal process that forms the core of the definition of affect. Therefore, affect, and the accompanying valence and arousal appraisals, should be considered an essential component for review in future empirical efforts with value.
Is affect biological? The literature has been somewhat inconsistent in exactly how affective states are made manifest in the brain so some review of what is meant by affective and biological together is mandated. The problem is stated quite well by Panksepp (1998) with, “…there are presently no direct metrics by which we can
unambiguously quantify changes in emotional states in any living creature” (p. 9). This is further elaborated by Murphy, Nimmo-Smith, and Lawrence (2003) who state, “Although some preliminary work has begun to show that anger, too, may be linked to activity of a distinct neural system, regions considered critical for the recognition of happiness, sadness, or surprise have not yet been identified” (p. 209).
This has not kept, however, research surrounding the idea that emotions have a specific biological basis from being attempted. In fact, more recent evidence, as presented by Cunningham, Bavel, and Johnsen (2008), suggests that evidence from multiple methodologies converges on a particular structure in the brain necessary for the
70
processing of affective properties of stimuli: the amygdala. Future researchers should use neuroimaging techniques to track amygdala activation in relation to efferent and afferent connections to other brain areas: particularly those areas related to judgment and decision making.
Despite the limitations presented for the stability element of the current research, some support is now provided overtly for the component of stability and value. The definition for stability presented in this research includes information on how stability is manifest in value as its own component, how it relates to the affective qualities of value, and how it ties to the possible physical, biological, structures of the brain that may instantiate value.
In summary of the previous work, stability represents the learned expressions of